Bluefish wind analysis

1 Introduction

This wind analysis was inspired by Munch and Conover (2000), who found a correlation between wind speed and coastwide YOY index. The wind measurements used are described below:

The NOAA environmental buoy array provided environmental data from 1970 to 1992. Buoys collected hourly records of windspeed and wind direction at a height of 10 m, and surface temperature at a depth of 1.5 m. Data were examined from a total of 34 buoys in fixed locations throughout the region from Cape Fear to the Gulf of Maine and extending from nearshore to beyond the shelf break. Buoy locations are reported in Munch (1997).

Spring-spawned YOY were apparently characterized as fish between 13-27cm in length in the autumn bottom trawl, but “A complete set of length-frequency distributions and cohort designations is presented in Munch (1997)”.

And the observed correlation is described:

A strong relationship existed between offshore wind and recruitment. Using north (n) and east (e) windstress components from NOAA buoys near the edge of the shelf of the Mid-Atlantic Bight, the model ln(c.p.u.e.) = -16.5 n - 12.8 e + 0.59 (p<0.05) explained 36% of the variance in spring cohort abundance in the SNE region

I am trying to obtain a copy of Munch (1997), which is a master’s thesis (Munch, S. B. 1997. Recruitment Dynamics of Bluefish, Pomatonus saltatrix, on the continental shelf from Cape Fear to Cape Cod, 1973–1995. M.S. Thesis. State University of New York at Stony Brook).

2 Determine relevant strata

2.1 What polygons to use?

Entire east coast:

Inshore (ish) strata:

All strata south of ~ Cape Cod

3 Daily data

3.1 Downloaded daily mean dataset (uwind, vwind) from NOAA PSL

Located in data-raw folder.

3.2 Extract mean wind by day

3.2.1 v wind

3.2.1.1 Read in wind data

# use ecopull function
devtools::install_github("kimberly-bastille/ecopull")
vwind <- ecopull::nc_to_raster(nc = here::here("data-raw/vwnd.10m.2021.nc"),
                               varname = "vwnd",
                               show_images = TRUE) 

vwind <- raster::rotate(vwind)
raster::plot(vwind)

3.2.1.2 Mask to relevant strata

m_vwind <- raster::mask(x = vwind, 
                        mask = new_shape[new_shape$STRATA > 2000 & 
                                           new_shape$STRATA < 8000,2])
mean(m_vwind@data@values, na.rm = TRUE)

## [1] -0.9522257

m_vwind[[1]]

## class      : RasterLayer 
## dimensions : 30, 20, 600  (nrow, ncol, ncell)
## resolution : 1.031519, 0.6498195  (x, y)
## extent     : -80.45845, -59.82808, 30.21661, 49.71119  (xmin, xmax, ymin, ymax)
## crs        : +proj=longlat +lat_0=40 +lon_0=-77 +x_0=0 +y_0=0 +datum=NAD83 +no_defs +ellps=GRS80 +towgs84=0,0,0 
## source     : memory
## names      : X2021.01.01 
## values     : 0.747654, 8.843357  (min, max)
## time       : 2021-01-01

raster::plot(m_vwind)

3.2.1.2.1 Plot over map

Seems like the existing strata are too small, try masking to more strata.

3.2.1.3 Mask to different strata

m_vwind2 <- raster::mask(x = vwind, 
                        mask = large_geom)
mean(m_vwind2@data@values, na.rm = TRUE)

## [1] -0.9731909

m_vwind2[[1]]

## class      : RasterLayer 
## dimensions : 30, 20, 600  (nrow, ncol, ncell)
## resolution : 1.031519, 0.6498195  (x, y)
## extent     : -80.45845, -59.82808, 30.21661, 49.71119  (xmin, xmax, ymin, ymax)
## crs        : +proj=longlat +lat_0=40 +lon_0=-77 +x_0=0 +y_0=0 +datum=NAD83 +no_defs +ellps=GRS80 +towgs84=0,0,0 
## source     : memory
## names      : X2021.01.01 
## values     : 0.1705055, 8.967381  (min, max)
## time       : 2021-01-01

raster::plot(m_vwind2)

3.2.1.3.1 Plot over map

This seems like a better geometry.

3.2.2 u wind

3.2.2.1 Read in wind data

# use ecopull function
devtools::install_github("kimberly-bastille/ecopull")
uwind <- ecopull::nc_to_raster(nc = here::here("data-raw/uwnd.10m.2021.nc"),
                               varname = "uwnd",
                               show_images = TRUE) 

uwind <- raster::rotate(uwind)
raster::plot(uwind)

3.2.2.2 Mask to strata

m_uwind2 <- raster::mask(x = uwind, 
                        mask = large_geom)
mean(m_uwind2@data@values, na.rm = TRUE)

## [1] 1.321539

m_uwind2[[1]]

## class      : RasterLayer 
## dimensions : 30, 20, 600  (nrow, ncol, ncell)
## resolution : 1.031519, 0.6498195  (x, y)
## extent     : -80.45845, -59.82808, 30.21661, 49.71119  (xmin, xmax, ymin, ymax)
## crs        : +proj=longlat +lat_0=40 +lon_0=-77 +x_0=0 +y_0=0 +datum=NAD83 +no_defs +ellps=GRS80 +towgs84=0,0,0 
## source     : memory
## names      : X2021.01.01 
## values     : 1.038357, 7.00613  (min, max)
## time       : 2021-01-01

raster::plot(m_uwind2)

3.2.2.2.1 Plot over map

3.2.3 Calculate weekly mean u and v wind

3.2.3.1 v wind

Break into north and south regions at 39N (approximately the bottom of NJ)

Positive v wind = wind is blowing to the north

##         names         mean_wind            min_wind          max_wind
## 1 X2021.01.01  8.67782974243164    8.32382583618164  8.96738052368164
## 2 X2021.01.02 0.727199292182922 -0.0607889890670776  1.43823444843292
## 3 X2021.01.03    2.053333568573      1.762610912323    2.250892162323
## 4 X2021.01.04 0.892561662197113   -1.19894230365753  2.63406562805176
## 5 X2021.01.05 -4.73075468540192   -6.87157487869263 -2.68505167961121
## 6 X2021.01.06  2.14590229988098    1.55810940265656  2.86865615844727

##         names          mean_wind          min_wind            max_wind
## 1 X2021.01.01   4.19205500284831 0.170505523681641    8.33652114868164
## 2 X2021.01.02  -2.03624478975932 -3.27465629577637 0.00757038593292236
## 3 X2021.01.03   1.81625674565633 0.796790599822998      2.449134349823
## 4 X2021.01.04   3.93146146138509  1.99344062805176    5.49441719055176
## 5 X2021.01.05  -1.92808553377787 -3.53368449211121  -0.494621992111206
## 6 X2021.01.06 -0.980953073501587 -5.24364852905273    1.50732815265656

3.2.3.2 u wind

Positive u wind = wind is blowing to the east

##         names         mean_wind           min_wind          max_wind
## 1 X2021.01.01  1.62507553100586   1.03835678100586  2.60866928100586
## 2 X2021.01.02  13.3995737075806   12.3858041763306  14.2393198013306
## 3 X2021.01.03  4.90297966003418   3.98139762878418  5.40229606628418
## 4 X2021.01.04  1.18881988525391 -0.510887145996094  2.53403472900391
## 5 X2021.01.05 -5.39001960754395  -6.41169929504395 -4.32380867004395
## 6 X2021.01.06 -10.1598304748535  -10.3247718811035 -9.93317031860352

##         names         mean_wind          min_wind          max_wind
## 1 X2021.01.01  5.45678126017253  2.56863021850586  7.00613021850586
## 2 X2021.01.02  11.8261691411336  9.40728950500488  13.8184213638306
## 3 X2021.01.03  3.47338981628418  2.34663200378418  4.80463981628418
## 4 X2021.01.04  3.07706858317057  2.32114410400391  3.48325347900391
## 5 X2021.01.05 -3.09281908671061 -4.70564460754395 -1.17439460754395
## 6 X2021.01.06 -8.94026629130046 -10.4019203186035  -4.1235990524292

3.3 Calculate across and along shore winds by week

Angles/directions are probably not reliable yet - have to ground truth

week	region	v mean	u mean	wind magnitude	wind blowing to	coarse angle	coast angle	adjusted angle	new angle	new wind blowing to	longshore wind	crossshore wind	check magnitude
1	North	3.026	-0.357	3.047	northwest	1.453	0.785	1.688	2.474	up_towards	2.392	-1.887	TRUE
1	South	1.578	0.631	1.699	northeast	1.190	0.785	1.190	1.976	up_towards	0.670	-1.562	TRUE
2	North	-0.094	3.338	3.339	southeast	0.028	0.785	6.255	0.757	up_away	2.427	2.294	TRUE
2	South	-2.115	3.345	3.958	southeast	0.564	0.785	5.719	0.222	up_away	3.861	0.870	TRUE
3	North	2.150	-4.333	4.837	northwest	0.461	0.785	2.681	3.466	down_towards	-4.584	-1.544	TRUE
3	South	2.550	-1.892	3.175	northwest	0.932	0.785	2.209	2.995	up_towards	3.141	-0.465	TRUE

3.3.1 Try plotting to see if things look correct

4 Monthly data

4.1 Extract mean wind by month

4.1.1 v wind

4.1.1.1 Read in wind data

# use ecopull function
vwind <- ecopull::nc_to_raster(nc = here::here("data-raw/vwnd.10m.mon.mean.nc"),
                               varname = "vwnd",
                               show_images = TRUE) 

vwind <- raster::rotate(vwind)
raster::plot(vwind)

4.1.1.2 Mask to relevant strata

m_vwind <- raster::mask(x = vwind, 
                        mask = large_geom)
mean(m_vwind@data@values, na.rm = TRUE)

## [1] -0.1615039

m_vwind[[1]]

## class      : RasterLayer 
## dimensions : 30, 20, 600  (nrow, ncol, ncell)
## resolution : 1.031519, 0.6498195  (x, y)
## extent     : -80.45845, -59.82808, 30.21661, 49.71119  (xmin, xmax, ymin, ymax)
## crs        : +proj=longlat +lat_0=40 +lon_0=-77 +x_0=0 +y_0=0 +datum=NAD83 +no_defs +ellps=GRS80 +towgs84=0,0,0 
## source     : memory
## names      : X1979.01.01 
## values     : -0.188582, 2.251501  (min, max)
## time       : 1979-01-01

raster::plot(m_vwind)

4.1.1.2.1 Plot over map

4.1.2 u wind

4.1.2.1 Read in wind data

# use ecopull function
uwind <- ecopull::nc_to_raster(nc = here::here("data-raw/uwnd.10m.mon.mean.nc"),
                               varname = "uwnd",
                               show_images = TRUE) 

uwind <- raster::rotate(uwind)
raster::plot(uwind)

4.1.2.2 Mask to relevant strata

m_uwind <- raster::mask(x = uwind, 
                        mask = large_geom)
mean(m_uwind@data@values, na.rm = TRUE)

## [1] 2.585091

m_uwind[[1]]

## class      : RasterLayer 
## dimensions : 30, 20, 600  (nrow, ncol, ncell)
## resolution : 1.031519, 0.6498195  (x, y)
## extent     : -80.45845, -59.82808, 30.21661, 49.71119  (xmin, xmax, ymin, ymax)
## crs        : +proj=longlat +lat_0=40 +lon_0=-77 +x_0=0 +y_0=0 +datum=NAD83 +no_defs +ellps=GRS80 +towgs84=0,0,0 
## source     : memory
## names      : X1979.01.01 
## values     : 0.6709892, 1.908766  (min, max)
## time       : 1979-01-01

raster::plot(m_uwind)

4.1.2.2.1 Plot over map

4.1.3 Calculate monthly mean u and v wind

4.1.3.1 v wind

Break into north and south regions at 39N (approximately the bottom of NJ)

Positive v wind = wind is blowing to the north

##         names          mean_wind           min_wind           max_wind
## 1 X1979.01.01   2.03434200286865    1.6679892539978   2.25150108337402
## 2 X1979.02.01  -2.61082348823547  -3.20551514625549  -2.16903352737427
## 3 X1979.03.01   1.78748390674591   1.65858399868011   1.87497127056122
## 4 X1979.04.01  -2.72354888916016  -3.43132877349854  -1.65548253059387
## 5 X1979.05.01 -0.771568644046783 -0.905924737453461 -0.616295218467712
## 6 X1979.06.01   2.62528488636017   2.42658042907715   2.83055186271667

##         names          mean_wind           min_wind          max_wind
## 1 X1979.01.01  0.699965926508109  -0.18858203291893  1.67570722103119
## 2 X1979.02.01  -3.28942049344381   -3.6595470905304  -2.9193127155304
## 3 X1979.03.01   1.42110284566879  0.730975568294525  1.87005698680878
## 4 X1979.04.01  -3.04131368001302  -3.55112051963806 -1.95717525482178
## 5 X1979.05.01 -0.222826617552588 -0.614342093467712 0.191290482878685
## 6 X1979.06.01   3.07070374488831   2.58722496032715  3.56281089782715

4.1.3.1.1 Test correlation with recruitment index from the 2019 assessment model

4.1.3.2 u wind

Positive u wind = wind is blowing to the east

##         names          mean_wind          min_wind           max_wind
## 1 X1979.01.01   1.45269603729248  1.29819440841675   1.57005679607391
## 2 X1979.02.01   2.63577941656113  1.98740470409393   3.57330727577209
## 3 X1979.03.01   4.12896010875702  3.96217894554138   4.27426958084106
## 4 X1979.04.01 -0.956579941511154 -1.30267691612244 -0.527709543704987
## 5 X1979.05.01  0.362831687927246 0.299994617700577  0.479556113481522
## 6 X1979.06.01   3.60882782936096  3.23824834823608   3.94925093650818

##         names         mean_wind           min_wind          max_wind
## 1 X1979.01.01  1.38577417532603  0.670989215373993  1.90876638889313
## 2 X1979.02.01  3.35631279945374   1.98196387290955  5.00613355636597
## 3 X1979.03.01  2.57121164798737   1.23079669475555  3.95357894897461
## 4 X1979.04.01 0.616446284453074 -0.510847508907318  1.37352740764618
## 5 X1979.05.01 0.527386673291524  0.396705806255341 0.648249387741089
## 6 X1979.06.01  2.52984784444173   1.16936814785004  3.22574830055237

4.1.3.2.1 Test correlation with recruitment index from the 2019 assessment model

4.1.4 Rotate

4.1.4.1 Determine rotation angle

Angle from north: 0.7853982 (pi / 4)

4.1.4.2 Do the rotation

names	u mean	Year	DOY	week	region	month	v mean	wind magnitude	ns	ew	wind blowing to	coarse angle	coast angle	adjusted angle	new angle	new wind blowing to	longshore wind	crossshore wind	check magnitude
1979.01.01	1.453	1979	1979-01-01	1	North	1	2.034	2.500	north	east	northeast	0.951	0.785	0.951	1.736	up_towards	0.411	-2.466	TRUE
1979.02.01	2.636	1979	1979-02-01	5	North	2	-2.611	3.710	south	east	southeast	0.781	0.785	5.503	0.005	up_away	3.710	0.018	TRUE
1979.03.01	4.129	1979	1979-03-01	9	North	3	1.787	4.499	north	east	northeast	0.408	0.785	0.408	1.194	up_away	1.656	4.183	TRUE
1979.04.01	-0.957	1979	1979-04-01	13	North	4	-2.724	2.887	south	west	southwest	1.233	0.785	4.375	5.160	down_away	-1.249	2.603	FALSE
1979.05.01	0.363	1979	1979-05-01	18	North	5	-0.772	0.853	south	east	southeast	1.131	0.785	5.152	5.937	down_away	-0.803	0.289	TRUE
1979.06.01	3.609	1979	1979-06-01	22	North	6	2.625	4.463	north	east	northeast	0.629	0.785	0.629	1.414	up_away	0.696	4.408	TRUE

Alongshore wind (positive = going up the coast)

Across shore wind (positive = going away from the coast)

4.1.4.2.1 Test correlation with recruitment index from the 2019 assessment model

4.1.4.2.2 Try GLMs

Poisson:

## 
## Call:
## glm(formula = Recruitment ~ 1, family = "poisson", data = dat)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -2827.9  -1506.9   -333.6    856.8   6709.6  
## 
## Coefficients:
##              Estimate Std. Error z value Pr(>|z|)    
## (Intercept) 1.765e+01  2.558e-05  690127   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for poisson family taken to be 1)
## 
##     Null deviance: 127825358  on 32  degrees of freedom
## Residual deviance: 127825358  on 32  degrees of freedom
## AIC: 127826002
## 
## Number of Fisher Scoring iterations: 4

## 
## Call:
## glm(formula = eqn, family = "poisson", data = dat)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -3061.4  -1022.8   -399.2   1038.8   5513.3  
## 
## Coefficients:
##                               Estimate Std. Error   z value Pr(>|z|)    
## (Intercept)                  1.771e+01  4.528e-05 391177.00   <2e-16 ***
## crossshore_wind_April_North  2.970e-02  2.195e-05   1352.95   <2e-16 ***
## crossshore_wind_May_North    2.938e-02  2.281e-05   1288.39   <2e-16 ***
## crossshore_wind_April_South -7.030e-02  3.502e-05  -2007.54   <2e-16 ***
## crossshore_wind_May_South   -8.964e-04  2.867e-05    -31.26   <2e-16 ***
## longshore_wind_April_North  -1.889e-02  1.606e-05  -1176.00   <2e-16 ***
## longshore_wind_May_North     5.135e-02  2.525e-05   2033.39   <2e-16 ***
## longshore_wind_April_South  -3.122e-02  1.904e-05  -1639.45   <2e-16 ***
## longshore_wind_May_South    -5.242e-02  2.906e-05  -1804.21   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for poisson family taken to be 1)
## 
##     Null deviance: 127825358  on 32  degrees of freedom
## Residual deviance: 102120229  on 24  degrees of freedom
## AIC: 102120889
## 
## Number of Fisher Scoring iterations: 4

##      df       AIC
## mod1  9 102120889
## mod0  1 127826002

Negative binomial:

##  Family: nbinom2  ( log )
## Formula:          Recruitment ~ 1
## Data: dat
## 
##      AIC      BIC   logLik deviance df.resid 
##   1176.8   1179.8   -586.4   1172.8       31 
## 
## 
## Dispersion parameter for nbinom2 family (): 12.8 
## 
## Conditional model:
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept) 17.65115    0.04869   362.5   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

##  Family: nbinom2  ( log )
## Formula:          
## Recruitment ~ crossshore_wind_April_North + crossshore_wind_May_North +  
##     crossshore_wind_April_South + crossshore_wind_May_South +  
##     longshore_wind_April_North + longshore_wind_May_North + longshore_wind_April_South +  
##     longshore_wind_May_South
## Data: dat
## 
##      AIC      BIC   logLik deviance df.resid 
##   1185.0   1200.0   -582.5   1165.0       23 
## 
## 
## Dispersion parameter for nbinom2 family (): 16.1 
## 
## Conditional model:
##                              Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                 17.718932   0.078949  224.44   <2e-16 ***
## crossshore_wind_April_North  0.035028   0.037882    0.92    0.355    
## crossshore_wind_May_North    0.021936   0.036552    0.60    0.548    
## crossshore_wind_April_South -0.072945   0.060861   -1.20    0.231    
## crossshore_wind_May_South    0.001085   0.044455    0.02    0.981    
## longshore_wind_April_North  -0.023728   0.028939   -0.82    0.412    
## longshore_wind_May_North     0.046580   0.041739    1.12    0.264    
## longshore_wind_April_South  -0.025401   0.033836   -0.75    0.453    
## longshore_wind_May_South    -0.044873   0.046434   -0.97    0.334    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

##      df      AIC
## mod0  2 1176.787
## mod1 10 1185.029